Dynamic Elite Individual Setting Based Heterogeneous Comprehensive Learning Particle Swarm Optimization

Classical particle swarm optimization suffers from premature convergence. In this paper, the dynamic elite individual setting based heterogeneous comprehensive learning particle swarm optimization is proposed to overcome this drawback. First, the comprehensive learning mechanism and population partition mechanism is introduced to dynamically select and retain the excellent attributes of the best particle in each iteration. Then, the autonomy of an exploitation particle is randomly increased to improve the convergence accuracy and break out of local optima. Finally, CEC2017 benchmark functions are utilized to test the efficiency of the designed algorithm through comparison against six variants based on particle swarm optimization.